Figure 4

Example of the two regression model used for the analysis of covariance showing the change in serum protein G2 glycan levels following AA supplementation (P<0.001). For each glycan structure, two regression models were fitted. The first ignored AA and fitted a straight-line (solid blue line) relationship with time. The second allowed for different regression lines in the pre-supplement and supplement phase (solid black lines; coefficient −0.501 and 0.018, respectively). This was achieved by creating two time variables; the first is time up to AA, but is equal to zero afterwards. The second time variable is time after AA, but is equal to zero before AA.